Currently in os_linux.cpp os::dll_load in case of loading errors of a lib the elf header is examined to get the reason of the error and produce a good error message.
However currently the coding has 2 issues :
- in case the endianness of the loaded lib and the host architecture do not match, the LE <-> BE conversion of e_machine is missing.
- the error message could show the real elf architecture name in mismatch cases

example when loading a x86_64 lib on linux s390x (big-endian platform) - the elf_head.e_machine value is currently read without conversion
elf_head.e_machine: 15872
elf_head.e_machine converted: 62 (this is the correct code from elf.h for x86_64/amd64 libs)

Adjusted error message looks like this :
example from renaissance db-shootout benchmark (runs on linuxs390x and attempts to load an x86_64 JNI lib "by mistake")
message without the change :
 .... ELF file data encoding not big-endian (Possible cause: endianness mismatch)]

with change :
 .... ELF file data encoding not big-endian (Possible cause: can't load AMD 64 .so on a IBM System/390 platform)